BuzzCut (The Erosion Clipper)

[bmanic]

"It doesn't stay properly linked." ? What DAW and OS if you please? My god I thought everything was working when I put this out.... there's another one for the list. YAY

Reaper (latest version as of this writing) on Mac OS 14.6.1. Running the plugin as a VST3.

I can send you a short video clip of the behavior if you want.

[pekbro]

We need some kind of bmanic signal, so we can alert you when there's some sort of interesting controversy going on like this.

[legendCNCD]

I'm intriqued to test this. Unfortunately in the middle of moving, audio computers somewhere in the piles..
.. I'll be back.

[onerob]

Tangentially, on the topic of noise and truncation:

[LusiD_Music_UK]

Tangentially, on the topic of noise and truncation:

Yes! Love this. A photoshop trick I love is if you're working with something quite low resolution. You can often improve it's perceived fidelity by adding grain/noise to it. The idea is that the brain prefers random noise over clunky digital artifacts. Haven't watched the video yet! (watching now) But Yes!

[LusiD_Music_UK]

We need some kind of bmanic signal, so we can alert you when there's some sort of interesting controversy going on like this.

I imagine with AI you'd be able to create some sort of constantly scan the internet for "Juicy controversy on XYZ topic I like" alert. If that doesn't exist yet. It should... and probably will soon

[LusiD_Music_UK]

I'm intriqued to test this. Unfortunately in the middle of moving, audio computers somewhere in the piles..
.. I'll be back.

Yessssss. Please make me successful quick guys before one of the big vst companies yoinks my idea!

[LusiD_Music_UK]

"It doesn't stay properly linked." ? What DAW and OS if you please? My god I thought everything was working when I put this out.... there's another one for the list. YAY

Reaper (latest version as of this writing) on Mac OS 14.6.1. Running the plugin as a VST3.

I can send you a short video clip of the behavior if you want.

No worries! Strange. I'll stick that on the list with the other 10,000 things

[onerob]

Tangentially, on the topic of noise and truncation:

Yes! Love this. A photoshop trick I love is if you're working with something quite low resolution. You can often improve it's perceived fidelity by adding grain/noise to it. The idea is that the brain prefers random noise over clunky digital artifacts. Haven't watched the video yet! (watching now) But Yes!

Looking forward to a 'Blue Noise' edition of BuzzCut

[LusiD_Music_UK]

Looking forward to a 'Blue Noise' edition of BuzzCut

That's super interesting. When looking at the image representation, blue noise looked like slightly "ordered" higher frequency noise. When applied to audio, I actually have the tilt function for this reason. The noise tilts from (roughly) "pink - white - BLUE - violet". So I wonder how practically close that is to being the same thing already?.... what's also interesting is that I do tend to prefer the sound of the modulation when tilted slightly brighter than white (roughly blue) coincidence???

Check out 14:00 haha

[onerob]

I wonder, does tilting white noise towards a lower or higher frequency range have the effect of de-randomization? Or is some kind of not-fully-random noise signal needed to begin with, in order to achieve the effect of blue noise from the video?

[LusiD_Music_UK]

I wonder, does tilting white noise towards a lower or higher frequency range have the effect of de-randomization? Or is some kind of not-fully-random noise signal needed to begin with, in order to achieve the effect of blue noise from the video?

Well yes, true random is white noise. Higher or lower frequency tilted (and filtered) "pure random" is thus necessarily no longer "true random".
I'm leaning towards tilted spectrum noise being (at least practically) the same technique he's describing when applied to audio. Perhaps someone more educated on the topic could correct me on that...

[LusiD_Music_UK]

Here's what Claude had to say on the topic:

Does tilting white noise achieve the same thing?
Partially, and here's the nuance. When you high-pass filter or tilt white noise toward blue/violet, you ARE introducing temporal correlation between samples — the filter creates "memory," so consecutive samples become negatively correlated. In that sense, you genuinely ARE de-randomizing it, and this is practically very similar to the perceptual goal described in the video.
However, true blue noise as a generation algorithm enforces this property more rigorously than a simple spectral tilt would. A tilt is an approximation that gets you most of the perceptual benefit.
For audio specifically, the good news
This is essentially what noise shaping in professional dithering (POW-r, etc.) already does — it pushes quantization error energy toward higher frequencies where hearing is less sensitive. So LusiD's intuition is correct: the tilt function is already doing the conceptually equivalent thing for audio. The preference for slightly-blue noise over white when modulating probably isn't a coincidence at all.

[LusiD_Music_UK]

This is too juicy for marketing not to share...

BuzzCut’s erosion clipper is doing the exact same perceptual trick in the amplitude domain:

A normal hard clipper = deterministic math → inharmonic foldback aliasing, metallic ringing, “digital harshness.”
BuzzCut = stochastic threshold modulated by shaped noise → the clipping edge becomes probabilistic and “eroded.” The deterministic distortion is shattered into broadband noise that the ear treats as pleasant texture instead of error. It’s literally dithering the clip point itself.
That’s why users keep saying it sounds “analog,” “airier,” “more 3D,” and why you can push it 1 dB louder without the usual brittle collapse. You’re not just clipping; you’re dithering the truncation/clipping process in real time, exactly like the video’s noise-over-quantization trick.

[onerob]

I need an energy drink version – shatters my quantization error energy and tilts me towards a higher frequency...